Modelling phenolic and volatile composition to characterize the effects of pre-fermentative cold soaking in Tempranillo wines

Modelling phenolic and volatile composition to characterize the effects of pre-fermentative cold soaking in Tempranillo wines

Accepted Manuscript Modelling phenolic and volatile composition to characterize the effects of prefermentative cold soaking in Tempranillo wines Dr. J...

1MB Sizes 0 Downloads 57 Views

Accepted Manuscript Modelling phenolic and volatile composition to characterize the effects of prefermentative cold soaking in Tempranillo wines Dr. José L. Aleixandre-Tudó, Inmaculada Álvarez, Victoria Lizama, Hélène Nieuwoudt, María J. García, José L. Aleixandre, Wessel J. du Toit PII:

S0023-6438(15)30251-6

DOI:

10.1016/j.lwt.2015.10.033

Reference:

YFSTL 5028

To appear in:

LWT - Food Science and Technology

Received Date: 13 July 2015 Revised Date:

17 September 2015

Accepted Date: 12 October 2015

Please cite this article as: Aleixandre-Tudó, J.L, Álvarez, I., Lizama, V., Nieuwoudt, H., García, M.J., Aleixandre, J.L., du Toit, W.J., Modelling phenolic and volatile composition to characterize the effects of pre-fermentative cold soaking in Tempranillo wines, LWT - Food Science and Technology (2015), doi: 10.1016/j.lwt.2015.10.033. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1 2

Modelling phenolic and volatile composition to characterize the effects of pre-fermentative cold soaking in Tempranillo wines

3

5

José L. Aleixandre-Tudó,1,3,* Inmaculada Álvarez,3 Victoria Lizama,3 Hélène

RI PT

4

Nieuwoudt,2 María J. García,3 José L. Aleixandre3 and Wessel J. du Toit1,2

6

(1) Department of Viticulture and Oenology and (2)Institute for Wine

8

Biotechnology, Stellenbosch University, Private Bag X1, Matieland, 7602,

9

South Africa.

11

M AN U

10

SC

7

(3) Departamento de Tecnología de Alimentos, Universitat Politecnica de Valencia, Camino de Vera s/n, 46022, Valencia, Spain. *Corresponding author

13

Dr. Jose Luis Aleixandre- Tudó

14

Tel.: (+27) 21-808 9206

15

Fax: (+27) 21-808 4781

16

E-mail: [email protected]

18

EP

AC C

17

TE D

12

19 20 21

1

ACCEPTED MANUSCRIPT ABSTRACT

23

The impact of pre-fermentative cold soak, alone or in combination with dry ice

24

addition, on colour, phenolic and volatile composition of Tempranillo wines at 12

25

months after bottling was studied. A control wine without cold soak was also evaluated.

26

A sample set consisting of 66 wines was investigated. The results from ANOVA and

27

PCA analysis showed significant treatment-related differences for a number of chemical

28

measurements, as well as overlapping effects. Orthogonal projections to latent

29

structures discriminant analysis (OPLS-DA) of the data showed that the dry ice addition

30

treatment had a major effect on the anthocyanin fraction and on the levels of ethyl

31

decanoate, 2-phenylethyl acetate and decanoic acid. In comparison, the cold soak

32

treatment only had a slight effect on the bisulphite bleaching anthocyanins and volatile

33

composition.

M AN U

SC

RI PT

22

Keywords: cold soak, dry ice, phenolics, volatiles and OPLS-DA

35

1. INTRODUCTION

36

Polyphenols and volatiles are essential for wine colour, mouth-feel and flavour.

37

The relative proportions of anthocyanins and tannins can be adjusted during skin

38

maceration to achieve a wine capable of undergoing good evolution during aging

39

(Glories & Galvin, 1990). Therefore, when fruity and full bodied red wines are required,

40

pre-fermentative cold maceration appears as an alternative for winemakers (Cai et al.,

41

2014). However, the effects of pre-fermentative techniques are highly dependent on

42

grape sanitary status and phenolic ripeness (Álvarez, Aleixandre, García, & Lizama,

43

2006), as well as on the aromatic nature of the variety (Moreno-Pérez, Vila-López, &

44

Fernández-Fernández, 2013).

AC C

EP

TE D

34

2

ACCEPTED MANUSCRIPT Skin polyphenols and volatile compounds are extracted throughout the pre-

46

fermentative cold soak in the absence of ethanol (Gomez-Míguez, Gonzalez-Miret, &

47

Heredia, 2007). Dry ice (solid carbon dioxide) addition appears as a common method to

48

obtain the cold temperatures required for this technique (Heredia et al., 2010). After

49

crushing and dry ice addition, grape skin cells are broken and disorganized through

50

freezing, which facilitates aroma and phenolic extraction (Álvarez, Aleixandre, García,

51

Lizama, & Aleixandre-Tudó, 2009).

RI PT

45

Contradictory results have been observed in the literature regarding the effect of

53

pre-fermentative cold soak techniques on chemical wine composition. Several studies

54

have shown that cold soak has either no effect, while a decrease in phenolic levels was

55

also observed (Heatherbell, Dicey, Goldsworthy, & Vanhanen, 1997; Marais, 2003;

56

Okubo, Goto-Yamamoto, & Okazaki, 2003: de Beer, Joubert, Marais, & Manley, 2006).

57

Other authors noticed an increased phenolic content and higher sensory scores when

58

wines were cold soaked before fermentation (Koyama, Goto-Yamamoto, & Hashizume,

59

2007; Gil-Muñoz et al., 2009; Gordillo, López-Infante, Ramírez-Pérez, González-Miret,

60

& Heredia, 2010: Heredia et al., 2010). During the cold soak step numerous reactions

61

between grape phenolics also occur, which may influence wine sensory properties

62

(Parenti, Spugnoli, Calamai, Ferrari, & Gori, 2004). Moreover, other compounds

63

(proteins and polysaccharides) are also extracted, which may participate in condensation

64

reactions (Gómez-Plaza, Gil-Muñoz, López-Roca, Martínez-Cutillas, & Fernández-

65

Fernández, 2001).

AC C

EP

TE D

M AN U

SC

52

66

Pre-fermentative techniques have been used extensively in the production of

67

white and rosé wines and have been recommended as a means for enhancing wine

68

aroma (Sanchez-Palomo, Gonzalez-Viñas, Díaz-Maroto, Soriano-Perez, & Perez-Coello

69

2007). Cold soak has also been tested in red winemaking (Sacchi, Bisson, & Adam, 3

ACCEPTED MANUSCRIPT 2005). However, little is known about its effect on the volatile composition. Moreno-

71

Perez, Vila-López, and Fernández-Fernández (2013) cited differences between cold

72

soak treated red wines and conventional winemaking after six months of bottling,

73

although no differences between pre-treatments (cold soak, freezing grapes and dry ice

74

addition) were observed. Moreover, in Monastrell wines increased volatile compound

75

levels were reported for Monastrell wines produced by dry ice addition after 6 months

76

of bottle storage (Álvarez, Aleixandre, García, & Lizama, 2006).

RI PT

70

Based on these contradictory reports, the aim of this work was thus to evaluate

78

the impact of cold soak techniques on color, phenolic and volatile composition of red

79

Tempranillo wines at 12 months of bottle storage. Tempranillo is one of the most

80

widely planted grape cultivars in Spain and wines made from this cultivar are

81

increasingly being accepted in new world wine producing countries such as Australia

82

and the United States (Cynkar, Dambergs, Smith, & Cozzolino, 2010; USDA, 2014).

M AN U

SC

77

2. MATERIALS AND METHODS

84

2.1. Wine Samples. Tempranillo grapes from a commercial vineyard (Utiel-

85

Requena, Valencia, Spain) were harvested in 2008. At harvest, the grapes had

86

221.33±9.29 g/L of total sugar content, total acidity of 6.1±0.2 g/L as tartaric acid, pH

87

of 3.34±0.12 and potential alcohol of 13.07±0.64% vol. (N=3). Wines were produced at

88

an experimental wine production centre (Universitat Politècnica de València (UPV)). 40

89

kg of grapes were destemmed, crushed, mixed and divided into closed 50 L stainless

90

steel tanks. Potassium bisulphite was added at 100 mg/kg before fermentation.

91

Treatments consisted of traditional vinification wines (T-V) without cold soak (control);

92

cold soaked wines at 6 to 8ºC for four days (C-S); and dry ice addition (0 to 2ºC)

93

followed by cold soak at 6 to 8ºC for four days (D-I). Twenty-two small scale

AC C

EP

TE D

83

4

ACCEPTED MANUSCRIPT vinifications were made for each treatment. After the cold soak period wines were left

95

to warm up in the fermentation room and commercial yeasts were inoculated at 20 g/hL

96

(Saccharomyces cerevisiae strain EP 841, Agrovin, Spain). The highest temperature

97

during fermentation was 25ºC. Manual punching down was carried out twice a day. T-V

98

wines were left on the skins for 15 days to ensure that sugar levels were lower than 2

99

g/L. Following the same workflow, C-S and D-I wines were also left on the skins for 15

100

days after the four days cold soak treatment. After fermentation wines were pressed and

101

the first 5 L were mixed with 20 L free-run wine. Oenococcus oeni strain OE 104

102

(Agrovin, Spain) lactic acid bacteria was inoculated and malolactic fermentation (MLF)

103

was conducted at room temperature (~20 ºC). Potassium bisulphite was added at 50

104

mg/L before bottling. Wines were stored at room temperature (15±2oC) and cork

105

closures were used.

M AN U

SC

RI PT

94

2.2. Analytical methods. A UV-Visible JASCO V-530 spectrophotometer, and

107

a JASCO MD-2010 Plus high-performance liquid chromatography instrument coupled

108

with a diode array detector (DAD) (JASCO LC-Net II/ADC, Tokyo, Japan) were used

109

for phenolic measurements. All the spectrophotometric measurements were performed

110

in triplicate. Using the analytical methods described by Glories (1984) colour intensity,

111

hue, gelatin (astringency) and EtOH (tannin-polysaccharide molecules) indexes were

112

estimated. The Ribéreau-Gayon and Stonestreet (1965) method was used for the

113

determination of bisulphite bleached anthocyanins. Catechins were quantified using the

114

method reported by Sun, Ricardo Da Silva, and Spranger (1998). The modified version

115

of the MCP tannin assay reported by Mercurio, Dambergs, Herderich, and Smith (2007)

116

was used for tannin quantification. The method reported by Boulton (1996) was used to

117

analyze the contribution of the copigmented, free and polymeric anthocyanins to the

118

total wine colour. PVPP (anthocyanin-tannin complexes) and DMACH (tannin degree

AC C

EP

TE D

106

5

ACCEPTED MANUSCRIPT 119

of polymerization) indexes were calculated according to Vivas and Glories (1995). The

120

Folin-Ciocalteu index was determined using the method developed by Singleton and

121

Rossi (1965). HPLC was used to quantify individual phenolic compounds using the method

123

reported by Jensen, Blachez, Egebo and Meyer (2007). Gallic acid, (+)-catechin and (-)-

124

epicatechin were quantified at 280 nm. Flavan-3-ols were defined as the sum of (+)-

125

catechin and (-)-epicatechin. Hydroxycinnamic acids were quantified at 316 nm.

126

Phenolic acids were calculated as the sum of gallic and hydroxycinnamic acids.

127

Flavonols were quantified at 365 nm. Total anthocyanins were calculated as the sum of

128

anthocyanidin-3-glucosides and acylated anthocyanins (derivated anthocyanins). Within

129

each phenolic group, compounds were identified based on their intrinsic spectral

130

features and retention times. Commercially available standards were used to build the

131

calibration curves for phenolics quantifications: gallic acid (Fluka, Milwaukee, WI,

132

USA), (+)-catechin (Fluka, Milwaukee, WI, USA) for flavan-3-ols, caffeic acid (Fluka,

133

Milwaukee, WI, USA) for hydroxycinnamic acids, rutin (Sigma-Aldrich, St Louis, MO)

134

for flavonols and malvidine-3-glucoside (Sigma-Aldrich, St Louis, MO) for

135

anthocyanins. 20 µL of the wine sample were injected twice after centrifugation (5000

136

rpm) and filtration (0.45 µm membrane Millipore filter). Separation was carried out on a

137

Gemini NX (Phenomenex, Torrance, CA) 5 µm, 250 mm x 4.6 mm i.d. column at 40ºC.

138

Acetonitrile and o-phosphoric acid were used as solvents. Solvents composition and the

139

elution gradient were reported elsewhere (Jensen, Blachez, Egebo, & Meyer, 2007).

AC C

EP

TE D

M AN U

SC

RI PT

122

140

An Agilent gas chromatograph (GC) (Agilent Technologies, Waldbronn,

141

Germany) equipped with a split/splitless capillary injection port and flame ionization

142

detector (FID) was used for the analysis of the wine aroma composition. Separations

143

were performed on a ZB- WAX Plus column (50 m x 0.25 mm i.d., 0.25 µm film 6

ACCEPTED MANUSCRIPT thickness) from Phenomenex (Aschaffenburg, Germany). Duplicate injections were

145

performed using the following conditions: injector temperature, 250°C; detector

146

temperature, 300°C; carrier gas flow (N2), 1 mL/min. Injections were made in split

147

mode (split ratio, 1/60; sample size, 1 µl). The oven temperature was maintained at

148

40°C for 7 min, from 40 to 110°C at 4 °C/min, from 110 to 170°C at 10°C/min, and

149

then held for 10 min. The comparison of retention times with those of standard

150

compounds was used to identify volatile compounds. Preparation of the samples was

151

carried out following the method proposed by Hernanz, Heredia, Beltran, and

152

Recamales (1999). Twenty volatile compounds were quantified with 2-octanol as

153

internal standard.

M AN U

SC

RI PT

144

2.3. Statistical analysis. Statgraphics Plus 5.1 software was used for the

155

ANOVA treatment of the data. Principal component analysis (PCA) and orthogonal

156

projections to latent structures discriminant analysis (OPLS-DA) (Trygg & Wold, 2002)

157

was performed using SIMCA version 13.0.3 software (www.umetrics.com). Chemical

158

measurements were Pareto (1/√Standard Deviation) scaled. The advantages of Pareto

159

scaling have been defined as a means of reducing the impact of noise and artefacts in

160

the models (Wiklund et al., 2008).

162

EP

3. RESULTS AND DISCUSION

AC C

161

TE D

154

3.1. Effect of pre-fermentative maceration techniques on wine phenolic

163

composition. ANOVA results showed no differences between the control and pre-

164

fermentative treated wines in colour density and hue after 12 months of bottling (Table

165

1). Bisulphite bleaching anthocyanins showed higher values in cold soaked (C-S) and

166

dry ice added (D-I) wines. Higher levels of malvidine-3-monoglucoside, sum of

167

anthocyanidins and total anthocyanins were found in D-I wines. These results are in

7

ACCEPTED MANUSCRIPT accordance with those obtained by Álvarez, Aleixandre, García, & Lizama (2006) and

169

Gil-Muñoz et al. (2009) which indicate an increase in the anthocyanin concentrations

170

due to cold soak with dry ice addition. The higher anthocyanin derivates values

171

observed might support the idea that dry ice addition is capable of increasing not only

172

the free anthocyanin fraction. The extraction of other low molecular weight phenolics

173

that could react with free anthocyanins might give rise to more complex anthocyanin

174

moieties (Aleixandre-Tudó et al., 2013). Gil-Muñoz et al. (2009) also reported increased

175

phenolic extraction and an improvement on the chromatic characteristics in cold soaked

176

wines.

SC

RI PT

168

Pre-fermentative cold soak techniques have been reported to influence the

178

presence of phenolic compounds with high copigmentation potential (Álvarez,

179

Aleixandre, García, Lizama, & Aleixandre-Tudó, 2009; Heredia et al., 2010). Despite

180

the reported results no differences in the copigmentation fraction were observed.

181

Copigmented anthocyanins account for almost half of the observed colour in young red

182

wines and it is thought that anthocyanins may take part in future condensation reactions.

183

Moreover, it is also believed that at early stages of fermentation the anthocyanin-

184

copigment molecules can prevent anthocyanin oxidation, favouring further covalent

185

tannin-anthocyanin interaction (Boulton, 2001).

TE D

EP

AC C

186

M AN U

177

The colour resulting from the free non-copigmented anthocyanins and the

187

flavan-3-ol concentration was also higher in D-I wines. Parenti, Spugnoli, Calamai,

188

Ferrari, and Gori (2004) reported increased flavan-3-ol, tannin and anthocyanin

189

concentrations in cold soaked wines after bottling, which led to more anthocyanin-

190

tannin interactions. Contrarily to what was reported, anthocyanin-tannin interactions

191

(PVPP index and % of colour due to polymeric anthocyanins) were not increased in this

192

study at 12 months of bottling. 8

ACCEPTED MANUSCRIPT 3.2 Effect of pre-fermentative cold soak techniques on wine volatile

194

composition. The ANOVA results showed that pre-fermentative cold soak techniques

195

caused significant differences (Table 2) in fatty acids (octanoic and decanoic acid),

196

ethyl esters (ethyl hexanoate and ethyl decanoate), an ester (2-Phenylethyl acetate), a

197

volatile phenol (4-vinylphenol) and a lactone (γ-butyrolactone). The concentration of

198

octanoic acid was lower in C-S and D-I wines, while the opposite was found for

199

decanoic acid, although only significant for D-I wines. Moreover, no trends were

200

observed in the esters and ethyl esters levels found for the different treatments. The

201

levels of 4-vinylphenol were lower for D-I wines when compared to T-V and C-S

202

treatments. Finally γ-butyrolactone levels seem to be diminished by cold soak

203

treatments, with T-V wines having the highest levels of this compound. Moreno-Pérez,

204

Vila-López, and Fernández-Fernández (2013) were not able to discriminate between the

205

volatile composition of cold soaked Monastrell, Cabernet Sauvignon and Syrah cold

206

soaked wines (freezing grapes, dry ice and cold soak) and concluded that these effects

207

were variety dependent.

TE D

M AN U

SC

RI PT

193

The large within treatment variations (as indicated by high standard deviations),

209

probably caused by the large number of small scale fermentations, could possibly

210

explain why non-significant differences were observed for some compounds (Table 2).

211

Contrarily to univariate techniques such as ANOVA that compare only one variable at

212

the time, multivariate techniques include more than one statistical variable and include

213

the effect of all variables on the responses of interest. PCA and OPLS-DA were thus

214

performed to explore the factors that could contribute to the variation between

215

treatments.

AC C

EP

208

216

3.3. Multivariate data analysis. In order to investigate the relationship between

217

chemical composition and treatments in a more holistic approach, PCA was performed 9

ACCEPTED MANUSCRIPT on the 66 samples and 44 variables (24 phenolic measurements and 20 volatile

219

compounds). The bi-plot (Figure 1 (a)) showed that the first two principal components

220

captured 93.7 % of the explained variance (PC1 = 60.6 % and PC2 = 33.1 %) in the

221

dataset. Although some differences were shown in the ANOVA, no clear trend based on

222

treatments was observed in the PCA plot (i.e. no grouping of samples), thereby making

223

it difficult to understand which chemical compounds were correlated to the applied pre-

224

fermentative techniques.

RI PT

218

As has been reported by several authors, cold pre-fermentative techniques might

226

initially increase anthocyanin and phenolic extraction (Gil-Muñoz et al., 2009),

227

although the reactions that occur during storage could reduce or even negate this effect.

228

In a study reported by Ortega-Heras, Pérez-Magariño, and González-Sanjosé (2012) the

229

higher extraction of some phenolic compounds induced by cold soak was no longer

230

evident at the end of bottle aging (12 months). Other authors reported that the effect

231

might depend on the vintage, which could be related to the phenolic maturity of the

232

grapes (Álvarez, Aleixandre, García, & Lizama, 2006, de Beer, Joubert, Marais, &

233

Manley, 2006). The discrepancies in the abovementioned studies indicated that the

234

effect could also depend on the variety used (Gil-Muñoz et al., 2009; Bussé-Valverde et

235

al., 2010).

M AN U

TE D

EP

AC C

236

SC

225

With the aim of further investigating the differences between treatments, the

237

OPLS-DA technique was applied to the data. OPLS-DA removes variability in X

238

(chemical data) that is orthogonal to Y (treatments), by separating systematic variation

239

into predictive components and orthogonal components (Lee, Hong, & Lee, 2009).

240

Figure 1 (B) shows OPLS-DA bi-plot of pre-fermentative cold soaked (C-S and

241

D-I) and control wines (T-V) using the complete set of 66 wines and 44 variables

10

ACCEPTED MANUSCRIPT (chemical data). The model resulted in two predictive components and five orthogonal

243

components, with cross-validated predictive ability Q2(Y) = 52.3 %, total explained

244

variance R2(X) = 98.7 %, and variance related to treatment effect R2p(X) = 7 %.

245

Compared to the results obtained with PCA, OPLS-DA exhibited an improved

246

separation of cold soaked and control wines (Figure 1 (b)). C-S and D-I wines were

247

mostly located towards the negative end of PC1, and T-V wines towards the positive

248

end. Moreover, a differentiation between C-S and D-I wines was observed on PC2, with

249

C-S wines located towards the positive end, and D-I towards the negative end.

SC

RI PT

242

In order to further investigate the effect of cold soak treatments on the chemical

251

composition of the wines, only the statistically significant compounds identified in the

252

ANOVA were considered. In addition, pairwise comparisons of the chemical

253

composition of T-V, C-S wines and D-I wines were performed. Figure 2 shows the

254

cross-validated OPLS-DA scores plots (a.1, b.1 and c.1) and the S-plots (a.2, b.2 and

255

c.2) that were constructed on the pair-wise comparisons. S-plots are scatter plots of the

256

predictive values obtained by cross validation vs. the correlation vectors. The S-plot

257

often takes an “S” shape if the X variables are scaled to Pareto variance. X variables

258

located far out of the “S” wings combine high covariance and correlation and are

259

therefore important contributors to the differences between treatments.

TE D

EP

AC C

260

M AN U

250

The comparison of significant variables in T-V and C-S wines is shown in

261

Figure 2(a.1 and a.2), and a clear separation between samples of the different treatments

262

was not observed (Figure 2(a.1)). The volatile compounds ethyl decaoate (EDe), ethyl

263

hexanoate (EHe) and octanoic acid (OcA) appeared located on the lower edge of the “S”

264

wing and are therefore highly correlated with T-V wines. On the other hand, looking at

265

the opposite side of the S-plot, the volatile compounds decanoic acid (DeA) and 2-

266

phenylethyl acetate (2PA) together with the bisulphite bleaching anthocyanins (BBA) 11

ACCEPTED MANUSCRIPT 267

were important contributors to C-S wines (Figure 2(A.2)). Figure 2(b.1 and b.2) also show the cross-validated OPLS-DA scores plot and

269

the S-plot for the pair T-V vs. D-I. A better separation between samples was clearly

270

observed, which could be interpreted as bigger differences in chemical composition

271

between wines, than those observed between C-S and T-V tretments (Figure 2(b.1)). For

272

the D-I and T-V pair, the volatile compounds octanoic acid (OcA), ethyl hexanoate

273

(EHe), γ-butirolactone (Btl) and 4-vinilphenol (4VP) appeared to be correlated with T-V

274

wines, while ethyl decanoate (Ede), 2-phenylethyl acetate (2PA) and decanoic acid

275

(DeA), together with the parameters related with the anthocyanin fraction (malvidine

276

(Mal), anthocyanidins (Ant), anthocyanin derivates (ADe), total anthocyanins (TAn),

277

bisulfite bleaching anthocyanins (BBA) and % free anthocyanins (%FA)) and the

278

parameter flavan-3-ols (Fla) appeared highly related with dry D-I wines (Figure 2(b.2)).

279

Finally, the direct comparison between pre-fermentative techniques (C-S vs. D-

280

I) is shown in Figure 2(c.1 and c.2). A clear separation between treatments was also

281

observed (Figure 2(c.1)). These results suggested that the dry ice addition treatment (D-I

282

wines) had the biggest influence on the volatile and phenolic composition of

283

Tempranillo wines after 12 months of bottle aging. C-S wines also differed from T-V

284

wines, although the variation in phenolic and volatile composition was less pronounced.

285

The volatile compounds ethyl decanoate (EDe), 2-phenylethyl acetate (2PA) and

286

decanoic acid (DeA), together with some of the mentioned anthocyanin related

287

parameters, were strongly correlated with the D-I wines. On the other hand, volatile

288

compounds such as octanoic acid (OcA), ethyl hexanoate (EHe), γ-butirolactone (Btl)

289

and 4-vinilphenol (4VP) were correlated with C-S wines (Figure 2(C.2)).

290

AC C

EP

TE D

M AN U

SC

RI PT

268

3.4. OPLS-DA classification. Classification models were established using half

12

ACCEPTED MANUSCRIPT of the samples (N=33) for the calibration set, while the other 33 samples were used to

292

validate the models. The classification results are shown in Table 3. In order to

293

investigate which group of chemical compounds had the biggest influence on the

294

differences between cold soak treatments the parameters were grouped based on the

295

corresponding phenolic family and also on their aromatic chemical families (esters,

296

higher alcohols, volatile phenols, fatty acids and lactones). Therefore groups containing

297

anthocyanin (BBA, Del, Cya, Pet, Peo, Mal, Ant, Ade, ToA, %CA, %FA, %PA, PVI),

298

tannin (PVI, Cat, Fla, Tan, DMI, EtI, GeI), phenolic acids (PhA) and flavonol (Flo)

299

related parameters were tested. Two additional groups, one including the total phenolics

300

parameter (FoI) and another considering the two wine colour related parameters (CDe

301

and Hue) were also performed. Two groups including all the phenolic measurement and

302

all the volatile compounds were also considered. Finally the classification was again

303

performed taking into account all the variables included in the study (phenolic +

304

volatiles). This last model showed a validation accuracy of 87.88% in classifying the

305

cold soak techniques, thus highlighting the impact that pre-fermentative cooling

306

techniques have on wine chemical composition.

TE D

M AN U

SC

RI PT

291

Within phenolic compounds, anthocyanin related parameters were 60.61%

308

accurately classified in validation showing the effect of these techniques on the

309

anthocyanin fraction. Finally when all the phenolic measurements were included in the

310

model 54.55 % of the samples were accurately classified.

AC C

311

EP

307

Regarding volatile compounds, the ester group appeared as the group of

312

compounds with the better ability to classify samples, with 63.64% of validation

313

accuracy. Otherwise other groups of volatile compounds showed lower accuracy in

314

classifying Tempranillo wines (Table 3). An important increase in ester levels has also

315

been observed in cold soaked and dry ice added pre-fermentative macerated wines 13

ACCEPTED MANUSCRIPT (Table 2). Compounds such as ethyl decanoate (soap, floral, grape, fruity, fatty, and

317

pleasant) and 2-phenylethyl acetate (rose, honey, tobacco, fruity, flowery, pleasant)

318

were identified as important contributors of D-I wines aroma. In addition, when the

319

volatile compounds were all included in the prediction model, 81,82% of the samples

320

were correctly classified.

321

CONCLUSIONS

322

Dry ice addition led to significant effects on anthocyanins and flaval-3-ols

323

levels. However an increased formation of tannins and polymeric pigments was not

324

observed. Dry ice also influenced volatile composition with increased levels of ethyl

325

decanoate, 2-phenylethyl acetate and decanoic acid. Pre-fermentative cold soak caused

326

minor changes in the anthocyanins and volatile fraction. An extended cold soak period

327

might cause larger effects, but in this case economic reasons need to be considered.

328

Moreover from the classification analysis cold soak treatments seem to strongly

329

influence the volatile composition, especially esters, together with the anthocyanin

330

fraction. Finally it is important to mention that these results should be confirmed with

331

further investigations which include at least three seasons due to our study only being

332

conducted on a single vintage.

334 335

SC

M AN U

TE D

EP

AKNOWLEDGMENTS

AC C

333

RI PT

316

The research reported here is a project financially supported by the Spanish

Government (AGL 2006-10723-C02-02) which the authors gratefully acknowledge.

336

REFERENCES

337

Aleixandre-Tudó, J. L., Álvarez, I., Lizama, V., García, M. J., Aleixandre, J. L.,

338

& du Toit, W. J. (2013). Impact of caffeic acid addition on phenolic composition of

14

ACCEPTED MANUSCRIPT 339

Tempranillo wines from different winemaking techniques. Journal of Agricultural and

340

Food Chemistry, 61, 11900-11912. Álvarez, I., Aleixandre, J. L., García, M. J., Lizama, V., & Aleixandre-Tudó, J.

342

L. (2009). Effect of the prefermentative addition of compigments on the polyphenolic

343

composition of Tempranillo wines after malolactic fermentation. European Food

344

Research and Technology, 228, 201−210.

RI PT

341

345

Álvarez, I., Aleixandre, J. L, García, M. J., & Lizama, V. (2006). Impact of

346

prefermentative maceration on the phenolic and volatile compounds in Monastrell red

347

wines. Analytica Chimica Acta, 563, 109−115.

Boulton, R. (2001). The copigmentation of anthocyanins and its role in the color

349

of red wine: A Critical Review. American Journal of Enology and Viticulture, 52, 67–

350

87.

M AN U

SC

348

351

Boulton, R. B. (1996). Methods for the assement of copigmentation in red wines.

352

Presented at the 47th Annual Meeting of the American Society for Enology and

353

Viticulture, Reno, NV.

Bussé-Valverde, N., Gómez-Plaza, E., López-Roca, J. M., Gil-Muñoz, R.,

355

Fernández-Fernández, J. I., & Bautista-Ortín, A. (2010). Effect of different enological

356

practices on skin and sees proanthocyanidins in three varietal wines. Journal of

357

Agricultural and Food Chemistry, 58, 11333-11339.

TE D

354

Cai, J., Zhu, B. Q., Wang, Y. H., Lu, L., Lan, Y. B., Reeves, M. J., & Duan, C.

359

Q. (2014). Influence of pre-fermentation cold maceration treatment on aroma

360

compounds of Cabernet Sauvignon wines fermented in different industrial scale

361

fermenters. Food Chemistry, 154, 217-229.

AC C

362

EP

358

Culleré, L., Escudero, A., Cacho, J., & Ferreira, V. (2004). Gas

363

chromatography– olfactometry and chemical quantitative study of the aroma of six

364

premium quality Spanish aged red wines. Journal of Agricultural and Food Chemistry,

365

52(6), 1653–1660.

366

Cynkar, W., Dambergs, R., Smith, P., & Cozzolino, D. (2010). Classification of

367

Tempranillo wines according to geographic origin: Combination of mass spectrometry

368

based electronic nose and chemometrics. Analytica Chimica Acta, 660(1–2), 227–231.

15

ACCEPTED MANUSCRIPT 369

De Beer, D., Joubert, E., Marais, J., & Manley, M. (2006). Maceration before

370

and during fermentation: effect on Pinotage wine phenolic composition, total

371

antioxidant capacity and objective colour parameters. South African Journal of Enology

372

and Viticulture, 27, 137-150. Ferreira, V., López, R., Cacho, J. F. (2000). Quantitative determination of the

374

odorants of young red wines from different grape varieties. Journal of the Science of

375

Food and Agriculture, 80, 1659–1667.

377

Francis, I. L., & Newton, J. L. (2005). Determining wine aroma from compositional data. Australian Journal of Grape and Wine Research, 11, 114–126.

SC

376

RI PT

373

Gil-Muñoz, R., Moreno-Pérez, A., Vila-López, R., Martínez-Cutillas, A.

379

Fernández-Fernández, J. I., & Gómez-Plaza, E. (2009). Influence of low temperature

380

prefermentative techniques on chromatic and phenolic characteristics of Syrah and

381

Cabernet sauvignon wines. European Food Research and Technology, 228, 777-788.

383 384 385

Glories, Y. (1984). La coleur des vins rouges. Mesure, origine et interpretation. Connaissance de la Vigne et du Vin, 18(4), 253-271.

Glories, Y, & Galvin C. (1990). Les complexes tanin-anthocyanes en présence

TE D

382

M AN U

378

d’éthanal. Conditions de leur formation. Acttualités œnologiques, 89, 408-413. Gómez-Míguez, M., Gonzalez-Miret, M., & Heredia, F. (2007). Evolution of

387

color and anthocyanin composition of Syrah wines elaborated with pre-fermentative

388

cold maceration. Journal of Food Engineering, 79, 271-278.

EP

386

Gómez-Plaza, E., Gil-Muñoz, R., López-Roca, J. M., Martínez-Cutillas, A., &

390

Fernández-Fernández, J. I. (2001). Phenolic compounds and color stability of red wines:

391

Effect of skin maceration time. American Journal of Enology and Viticulture, 52(3),

392

266-270.

393

AC C

389

Gordillo, B., López-Infante, M. I., Ramírez-Pérez, P., González-Miret, M., &

394

Heredia, F. (2010). Influence of prefermentative cold maceration on the color and

395

anthocyanin copigmentation of organic Tempranillo wines elaborated in warm climate.

396

Journal of Agricultural and Food Chemistry, 58, 6797-6803.

397

Heatherbell, D., Dicey, M., Goldsworthy, S., & Vanhanen, L. (1997). Effect of

398

prefermentation cold maceration on the composition, color and flavor of Pinot noir 16

ACCEPTED MANUSCRIPT 399

wine. Proceedings of the New Zealand Grape & Wine Symposium, G.F. Steans (Ed.),

400

30-42, New Zealand Society for Viticulture and Oenology, P.O. Box 90-276, Auckland

401

Mail Centre, Auckland 1, New Zealand. Heredia, F. J., Escudero-Gilete, M. L., Hernanz, D., Gordillo, B., Meléndez-

403

Martínez, A. J., Vicario, I. M., & González-Miret, M. L. (2010). Influence of the

404

refrigeration technique on the color and phenolic composition of Syrah wines obtained

405

with pre-fermentative cold maceration. Food Chemistry, 118, 377-383.

RI PT

402

Hernanz, D., Heredia, F. J., Beltran, R., & Recamales, M. A. F. (1999).

407

Optimization of an extraction method of aroma compounds in white wine using

408

ultrasound. Talanta, 50, 413–421.

409

SC

406

Jensen, J. S., Blachez, B., Egebo, M., & Meyer, A. S. (2007). Rapid Extraction of Polyphenols from Red Grapes. American Journal of Enology and Viticulture, 58 (4),

411

451-461.

M AN U

410

Koyama, K., Goto-Yamamoto, N., & Hashizume, K. (2007). Influence of

413

maceration temperature in red wine vinification on extraction of phenolics from berry

414

skins and seed of grape (Vitis Vinifera). Bioscience Biotechnology Biochemistry, 71(4),

415

958-965.

TE D

412

416

Lee, J. E., Hong, Y. S., & Lee, C. H. (2009). Characterization of fermentative

417

behaviors of lactic acid bacteria in grape wines through 1H NMR- and GC-based

418

metabolic profiling. Journal of Agricultural and Food Chemistry, 57, 4810–4817. Li, H., Tao, Y. S., Wang, H., & Zhang, L. (2008). Impact odorants of

420

Chardonnay dry white wine from Changli County (China). European Food Research

421

and Technology, 227 (1), 287–292.

AC C

422

EP

419

Marais, J. (2003). Effect of different wine-making techniques on the

423

composition and quality of Pinotage wine I. Low-temperature skin contact prior to

424

fermentation. South African Journal of Enology and Viticulture, 24, 70-75.

425

Mercurio, M. D., Dambergs, R.G., Herderich, M.J., & Smith, P.A. (2007). High

426

throughput analysis of red wine and grape phenolics: adaptation and validation of

427

methyl cellulose precipitable tannin assay and modified Somers color assay to a rapid

428

96 well plate format. Journal of Agricultural and Food Chemistry, 55(12), 4651-4657.

17

ACCEPTED MANUSCRIPT 429

Moreno-Pérez, A., Vila-López, R., & Fernández-Fernández, J. I. (2013).

430

Influence of cold pre-fermentation treatments on the major volatile compounds of three

431

wine varieties. Food Chemistry, 139(1–4), 770–776. Okubo, K., Goto-Yamamoto, N., & Okazaki, N. (2003). Effect of

433

prefermentative cold soak on extraction of anthocyanin during red wine making.

434

Journal of the Brewing Society of Japan, 98, 193-200.

RI PT

432

Ortega-Heras, M., Pérez-Magariño, S., & González-Sanjosé, M. (2012).

436

Comparative study of the use of maceration enzymes and cold prefermentative

437

maceration on phenolic and anthocyanic composition and color of a Mencía red wine.

438

LWT-Food Science and Technology, 48, 1-8.

SC

435

Parenti, A., Spugnoli, P., Calamai, L., Ferrari, S., & Gori, C. (2004). Effects of

440

cold maceration on red wine quality from Tuscan Sangiovese grape. European Food

441

Research and Technology, 218, 360-366.

442 443

M AN U

439

Ribéreau-Gayon, P., & Stronestreet, E. (1965). Le dosage des anthocyanes dans le vin rouge. Bulletin de la Société Chimique de France, 9, 119-142. Sacchi, K., Bisson, L., & Adams, D. (2005). A review of the effect of

445

winemaking techniques on phenolic interaction in red wines. American Journal of

446

Enology and Viticulture 53(3), 197-206.

TE D

444

Sánchez Palomo, E., González-Vinas, M. A., Díaz-Maroto, M. C., Soriano-

448

Pérez, A., & Pérez-Coello, M. S. (2007). Aroma potential of Albillo wines and effect of

449

skin-contact treatment. Food Chemistry, 103, 631–640.

EP

447

Singleton, V.L., & Rossi, J. (1965). A Colorimetry of total phenolics with

451

phosphomolybdic-phosphotungstic acid reagents. American Journal of Enology and

452

Viticulture, 16, 144-158.

453

AC C

450

Sun, B., Ricardo Da Silva, J.M., & Spranger, I. (1998). Critical factors of

454

vanillin assay for catechins and proanthocyanidins. Journal of Agricultural and Food

455

Chemistry, 46, 4267-4274.

456

Swiegers, J. H., Bartowsky, E. J., Henschke, P. A., & Pretorius, I. S. (2005).

457

Yeast and bacterial modulation of wine aroma and flavour. Australian Journal of Grape

458

and Wine Research, 11, 139-173.

18

ACCEPTED MANUSCRIPT 459 460 461 462

Trygg, J., & Wold, S. (2002). Orthogonal projections to latent structures (OPLS). Journal of Chemometrics, 16, 119–128. USDA (2014). National Agricultural Statistics Service. California Grape Crush report. Washington, DC. Vivas, N., & Glories, Y. (1995). Ranking of red wines matured in barrels. A

464

tentative classification of racking techniques. Australian and New Zealand Wine

465

Industry Journal, 10, 241-243.

RI PT

463

Wiklund, S., Johansson, E., Sjostrom, L., Mellerowicz, E.J., Edlund, U.,

467

Shockcor, J.P., Gottfries, J., Moritz, T., & Trygg, J. (2008). Visualization of GC/TOF-

468

MS-based metabolomics data for identification of biochemically interesting compounds

469

using OPLS class models. Analytical Chemistry, 80, 115-122.

SC

466

M AN U

470

AC C

EP

TE D

471

19

ACCEPTED MANUSCRIPT Table 1 Colour and phenolic profile of Tempranillo wines after 12 month of bottle aging. Mean ± standard deviation values of the phenolic measurement. p-value

12,71 ± 1.55

12,52 ± 0.88

12,21 ± 1.06

0,39

Hue

71,20 ± 2.63

69,93 ± 2.28

70,84 ± 2.28

0,20

Bisulfite bleaching ant. (mg/L) (3)

BBA

271,85 ± 27.83 a

297,59 ± 38.96 b

287,27 ± 30.64 b

0,04*

Delfinidin-3-glucoside (mg/L) (4)

Del

33,10 ± 2.84

32,21 ± 1.85

33,58 ± 1.73

0,19

Cya

2,17 ± 2.15

2,15 ± 0.25

Pet

30,67 ± 31.93

31,93 ± 1.90

Peonidin-3-glucoside (mg/L) (7)

Peo

5,13 ± 1.01

5,20 ± 0.87

Malvidin-3-glucoside (mg/L) (8)

Mal

255,42 ± 18.40 a

Anthocyanidins (mg/L) (9)

Ant

326,49 ± 24.64 a

Anthocyanin derivates (mg/L) (10)

ADe

46,84 ± 3.69 a

48,33 ± 7.28 a

53,34 ± 8.63 b

0,00*

Total anthocyanis (mg/L) (11)

ToA

373,33 ± 25.24 a

377,16 ± 34.48 a

407,78 ± 32.23 b

0,00*

% Copigmented anthocyanins (12)

%CA

10,48 ± 2.65

10,61 ± 2.39

10,48 ± 2.25

0,91

% Free anthocyanins (13)

%FA

39,49 ± 1.21 a

40,72 ± 2.91 ab

41,98 ± 2.76 b

0,04*

% Polimerized anthocyanins (14)

%PA

PVPP index (15)

PVI

Folin index (16)

FoI

Phenolic acids (mg/L) (17)

PhA

Flavonols (mg/L) (18)

Flo Cat Fla

Tannins (g/L) (21)

Tan DMI

0,65

0,08

5,49 ± 0.90

0,49

257,34 ± 26.61 a

280,43 ± 25.88 b

0,00*

328,83 ± 30.30 a

354,45 ± 29.36 b

0,00*

50,03 ± 7.62

48,67 ± 3.54

47,54 ± 4.70

0,77

32,02 ± 3.07

32,79 ± 2.90

33,92 ± 2.98

0,07

46,22 ± 2.23

45,65 ± 1.95

45,56 ± 1.95

0,50

146,74 ± 29.40

148,56 ± 26.35

137,98 ± 26.56

0,34

61,27 ± 20.59

57,19 ± 20.87

63,38 ± 19.34

0,59

M AN U

Catechin assay (mg/L) (19) Flavan-3-ols (mg/L) (20)

2,33 ± 0.03

32,61 ± 2.42

RI PT

Cyanidin-3-glucoside (mg/L) (5) Petunidin-3-glucoside (mg/L) (6)

SC

CDe

Ethanol index (23)

85,49 ± 17.59

84,93 ± 15.05

85,49 ± 13.88

0,99

65,45 ± 19.01 a

70,16 ± 20.00 ab

80,89 ± 18.34 b

0,03*

2,15 ± 0.25 b

2,01 ± 0.20 a

2,03 ± 0.23 a

0,03*

43,76 ± 3.21

44,82 ± 3.99

0,29

43,41 ± 4.23

EtI

21,75 ± 5.89

23,96 ± 4.39

22,78 ± 4.83

0,25

GeI

64,97 ± 5.96

65,45 ± 10.12

65,53 ± 8.65

0,96

EP

n: number of samples involved in the analysis from each treatment. T-V: traditional vinification; C-S: cold soak; D-I: dry ice.*Statistical analysis ANOVA at 95% confidence level with same letters indicating no significant difference for Tukey’s test

AC C

8

D-I (n=22)

Hue (2)

Gelatin index (24)

4 5 6 7

C-S (n=22)

Color density (1)

DMACH index (22)

3

T-V (n=22)

TE D

1 2

1

ACCEPTED MANUSCRIPT

T-V (n=22)

C-S (n=22)

D-I (n=22)

p-value

PTb,b

IsA

449,55 ± 221,45

704,91 ± 349,04

675,18 ± 550,59

0,07

30(I)#

Ethyl hexanoate (26)

EHe

173,27 ± 125,21 b

105,48 ± 34,87 a

85,90 ± 19,71 a

0,00*

50(I)#

Ethyl lactate (27)

ELa

12797,10 ± 10021,19

23283,53 ± 19508,42

21629,45 ± 20015,09

0,10

14000(II)¥

Cis-3-hexen-1-ol (28)

C3H

26,30 ± 19,55

18,22 ± 9,84

21,14 ± 17,22

0,25

400(III)α

Ethyl 3-hydroxybutyrate (29)

EHb

21,21 ± 11,84

21,79 ± 9,90

22,33 ± 15,42

0,96

20000(III)ρ

Isobutyric acid (30)

IbA

2731,74 ± 1398,11

2998,24 ± 1646,48

3032,73 ± 2053,87

0,82

2300(IV)g

4-Vinylphenol (31)

4Vp

27,29 ± 22,48 b

26,58 ± 18,69 b

8,43 ± 9,04 a

0,00*

180(III)§

Butyric acid (32)

BtA

6,60 ± 4,54

7,45 ± 4,78

7,27 ± 5,65

0,84

173(III)α

Ethyl decanoate (33)

EDe

425,89 ± 228,45 b

190,63 ± 68,13 a

573,95 ± 119,04 c

0,00*

200(I)¥

γ-Butyrolactone (34)

Btl

40,58 ± 23,75 b

26,77 ± 10,83 a

23,77 ± 9,69 a

0,00*

35(III)ρ

Isopentanoic acid (35)

IpA

3,69 ± 1,63

3,86 ± 2,44

Diethyl succinate (36)

DeS

535,60 ± 317,26

614,19 ± 299,84

2-Phenylethyl acetate (37)

2PA

163,38 ± 74,92 a

213,14 ± 50,23 a

Hexanoic acid (38)

HeA

3121,63 ± 1459,38

3161,95 ± 1632,91

2-Methoxyphenol (39)

2MP

197,77 ± 68,82

2-Phenylethanol (40)

2Pe

4-Ethylguaiacol (41)

SC

RI PT

Isoamyl acetate (25)

4,03 ± 2,68

0,89

30(III)α

713,82 ± 170,98

0,10

200000(III)¥

306,31 ± 140,36 b

0,00*

250(IV)#

2717,16 ± 1619,91

0,59

420(V)g

249,24 ± 77,78

247,83 ± 111,20

0,09

9.5(III)α

40792,84 ± 30714,08

47685,85 ± 29690,71

50851.00 ± 30873,27

0,54

14000(IV)¥

4Eg

283,54 ± 97,22

238,30 ± 64,57

209,23 ± 150,89

0,09

33(IV)#

Octanoic acid (42)

OcA

576,25 ± 259,34 b

422,5 ± 233,28 a

350,92 ± 185,84 a

0,01*

500(III)α

4-Ethylphenol (43)

4Ep

51,40 ± 19,30

39,36 ± 15,64

43,57 ± 21,57

0,11

440(IV)§

Decanoic acid (44)

DeA

308,73 ± 104,77 a

340,20 ± 66,34 ab

408,89 ± 170,14 b

0,03*

1000(III)α

M AN U

3 4 5 6

Table 2 Volatile composition of Tempranillo wines after 12 month of bottle aging. Mean ± standard deviation values (µg/L) of the quantified volatile compounds.

n: number of samples involved in the analysis from each treatment. T-V: traditional vinification; C-S: cold soak; D-I: dry ice. PT: perception threshold. *Statistical analysis ANOVA at 95% confidence level with same letters indicating no significant difference for Tukey’s test. a

TE D

1 2

The reference from which the value was obtained is given in parentheses: I (Swiegers, Bartowsky, Henschke, & Pretorius, 2005); II (Li, Tao, Wang, & Zhang, 2008); III (Culleré, Escudero, Cacho, & Ferreira, 2004); IV (Francis & Newton, 2005); V (Ferreira, López, Cacho, 2000).

11 12 13 14 15

Symbols correspond to different matrices from which the PT was obtained: 10% ethanol/water solution; ¥12% ethanol/water solution; §12% ethanol/water solution with 8 g/L glycerol and different salts; ρ10% ethanol/water solution with 5 g/L of tartaric acid at pH 3.2; α11% ethanol/water solution with 7 g/L glycerol and 5 g/L tartaric acid; g10% ethanol/water solution with 7 g/L glycerol at pH 3.0.

AC C

16

b

#

EP

7 8 9 10

1

ACCEPTED MANUSCRIPT

3 4 5

Table 3 Summary of orthogonal projection to latent structures discriminant analysis (OPLS-DA) classification results

Color Total phenolics Anthocyanins Tannins Phenolic acids Flavonols Phenolics Esters Higher alcohols Fatty acids Volatile phenols Lactones Volatile compounds Phenolics + Volatiles

1,2 16 3-15 15,19-24 17 18 1-24 25-27,29,33,36,37 28,40 30,32,35,38,42,44 31,39,41,43 34 25-44 1-44

RI PT

Parameters included*

SC

Model

Cold soak Calibration (%) Validation (%) N=33 N=33 27,27 39,39 39,39 39,39 60,61 63,64 39,39 45,45 33,33 36,36 27,27 48,48 54,55 69,70 63,64 84,85 42,42 42,42 42,42 45,45 48,48 66,67 42,42 45,45 81,82 100 87,88 100

M AN U

1 2

*Parameters included in each model, numbers correspond to Table 1 and 2. 33 random selected samples were included in the calibration set while the remaining 33 were used for model validation.

AC C

EP

TE D

6

1

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

a)

Figure 1 Principal component analysis bi-plot (PCA) (A) and orthogonal projections on latent structures discriminant analysis bi-plot (OPLS-DA) (B) of phenolic determinations and volatile compounds of wines elaborated with pre-fermentative cold maceration techniques. ♦T-V wines; ●C-S wines; ■D-I wines. Phenolic and volatile abbreviations correspond to Table 1 and 2

b)

b.1)

EP

c.1)

AC C

c.1)

TE D

M AN U

b.2)

SC

a.1)

RI PT

ACCEPTED MANUSCRIPT a.2)

Figure 2 OPLS-DA scores plots and S-Plots considering the significant phenolic and volatile parameters identified in the ANOVA of wines elaborated with pre-fermentative cold maceration techniques. Comparison among T-V and C-S wine (A.1 and A.2), T-V and D-I wines (B.1 and B.2) and C-S and D-I wines (C.1 and C.2). ♦T-V wines, ●C-S wines, ■D-I wines. Phenolic and volatile abbreviations correspond to Table 1 and 2.

ACCEPTED MANUSCRIPT HIGHLIGHTS The effect of two cold soak techniques was investigated against control wines. Tempranillo wines were analyzed after twelve months of bottle storage

RI PT

Dry ice addition caused an important effect on wine anthocyanins and volatiles Differences between cold soak techniques were also observed

AC C

EP

TE D

M AN U

SC

The length of cold soak vs dry ice addition needs to be considered